OBJECTIVE: A data-driven approach is needed to evaluate the impact of disease-inclusion criteria for expanded carrier screening (ECS) panels as recommended in clinical guidelines.1 Such criteria suggest that each screened condition have a carrier frequency of greater than 1 in 100, cause a severe phenotype, and be amenable to prenatal diagnosis. The manner in which these criteria are interpreted and implemented could have an appreciable effect on the ability of ECS to detect couples at-risk for conceptuses with serious diseases.
DESIGN: De-identified data from 44,483 patients who underwent ECS on a 176-condition panel were analyzed to assess criteria suggested in clinical guidelines for inclusion of diseases on ECS panels.
MATERIALS AND METHODS: Seven clinical criteria suggested by the American College of Obstetricians and Gynecologists (ACOG)1 were evaluated by determining how exclusion of conditions that did not meet these criteria impacted the rates of detection of both individual carriers and at-risk couples. Assuming that a disease is worthy of screening only if the clinical sensitivity is high, we introduce a statistical framework that estimates the achievable detection rate of carriers for rare diseases by modeling whether there are enough reported cases to interpret the pathogenicity of observed variants.
RESULTS: 170 (97%) of the 176 conditions met at least six of the ACOG criteria,1 including those relating to disease severity and availability of prenatal diagnosis. Limiting an ECS panel to conditions meeting these criteria would reduce identification of at-risk couples by 3%. The seventh criterion, that a disease must have a carrier frequency above 1-in-100, can be interpreted in several different ways that have important consequences, as carrier frequencies vary widely by ethnicity. Depending on the definition used, the criterion is satisfied by between 12 (7%) and 75 (43%) conditions that account for between 5% and 51% of at-risk couples, respectively. Our modeling suggests that even a disease with a carrier frequency of 1 in 4,000 would achieve a clinical sensitivity of 50%, making it a plausible candidate for inclusion on an ECS panel. We demonstrate that inclusion of such rare diseases does not appreciably increase patient-management burden because the panel-wide screen-positive rate has mostly saturated from the panel’s common diseases alone.
CONCLUSIONS: In order for an ECS panel to be clinically meaningful to patients, disease-inclusion criteria—such as those outlined in current ACOG guidelines1—are needed. However, by analyzing data from a large patient cohort, we demonstrate that the 1-in-100 carrier-frequency threshold definition is unclear and arbitrary. When strictly enforced, it limits at-risk couple detection without yielding a substantial reduction in the clinical management needed to implement ECS, suggesting that this threshold should be revisited. To that end, we propose an alternative model that can identify when a disease is too rare to include in an ECS panel based on its estimated clinical sensitivity.